Andrew ([info]perspectivism) wrote,
@ 2006-07-20 18:00:00
Previous Entry  Add to memories!  Share this!  Next Entry
"Time Enough For Counting"
Aaron Brown essentializes poker's game theory frontier in Willmott magazine (a journal of quantitative finance):

A minority of games have "vying" aspects, chances to adjust the stake with impact on the play of the game. Poker is unique as a pure vying game. The outcome is entirely luck. Skill cannot be exercised on the cards, only on the setting of the stake. [...]

[Vying] is the only theoretically interesting popular game. There is interesting work to be done on games like chess and go, but the basic theory is already known. Existing algorithms could play perfectly. But we still have no accepted theory for poker, or bidding in bridge, or Monopoly. [...]

I think the proper way to analyze poker is through derivative accounting. [...]

The only edge you get in poker [from someone who knows your full basic strategy] comes from bluffing. [...]

Poker may be the only activity popularly considered to evidence both masculinity and intelligence.
(nice pdf)

(5 comments) - (Post a new comment)

I don't know much about poker
[info]catithat
2006-07-26 01:47 am UTC (link)
I don't play poker often and am a horrible player, but I've been playing go quite a lot this summer, which has been fun, and which shifts my mind in that direction. I think a lot of the strategy in go also has vying at its core, in 2 ways.

First, each move's biggest effect is to raise the stakes in one part of the board (by the possibility of a second move nearby). In a close game, this tends towards brinksmanship, as each move tries to be as close as possible to being stupid, while still being good.

Secondly, though the game is a perfect-information game, an important skill (which I'm quite poor at) is assessing the stakes in different parts of the board, and the size/importance of possible moves.

(Reply to this) (Thread)
go
[info]perspectivism
2006-07-26 09:18 pm UTC (link)

Very interesting!

So the "vying" has to do with representing your perceptions/judgments rather than concrete hidden variables like a pair of aces.

Great expanded concept!

(Reply to this) (Parent)
Re: I don't know much about poker
[info]scottsch
2006-07-27 11:34 pm UTC (link)
While there is a lot of brinkmanship in well-played go, it is also important to remember that "go is a sharing game."

I find the quoted dismissal of go to be, frankly, gay. "Existing algorithms could play perfectly"? Let's see it, buddy!

(Reply to this) (Parent)(Thread)
Re: I don't know much about poker
(Anonymous)
2006-10-23 08:44 pm UTC (link)
I wrote the article and didn't intend the Go comment to be a dismissal. In principle, it's possible to search every possible Go game and find the perfect strategy. I'm aware that the job is far beyond the computational power of existing computers, so that someone is going to have to come up with a clever way of reducing the necessary computations if we want an answer anytime soon. But that cleverness has nothing to do with the game of Go, and everything to do with algorithms.

To put it another way, we can solve Go for smaller boards. The basic theory is there, it's the size of the game that makes computing the result of that theory intractable.

None of that makes Go a bad or trivial game. But poker is different because no one knows how to solve it with any amount of computer power. Even if you reduce it to the smallest non-trivial form, the solution is unknown.

To put it even a third way, a solution to Go would be a triumph for the field of computer science. If the techniques could be used for other problems, it would be useful. But it would be a breakthrough in technique, not in understanding. A solution to poker, would change the way we think about the game.

(Reply to this) (Parent)(Thread)
Re: I don't know much about poker
[info]perspectivism
2006-10-23 08:49 pm UTC (link)

I wrote the article

Very cool. (Hi, Aaron!)

Referrer logs are great!

(Reply to this) (Parent)

(5 comments) - (Post a new comment)

Create an Account
Forgot your login or password?
Log in with OpenID
English • Español • Deutsch • Русский…